seq_linter() gives the wrong message for n:1 lints

[MichaelChirico]

The result message should mention rev(), but currently it doesn't:

lint(text = ".N:1", linters = seq_linter())
# <text>:1:1: warning: [seq_linter] .N:1 is likely to be wrong in the empty edge case. Use seq_len(1) instead.
# .N:1
# ^~~~

It also says seq_len(1) instead of seq_len(.N)

[IndrajeetPatil]

The same issue also holds for all other relevant functions I think:

library(lintr)

lint(text = "length(x):1", linters = seq_linter())
#> <text>:1:1: warning: [seq_linter] length(...):1 is likely to be wrong in the empty edge case. Use seq_along(...) instead.
#> length(x):1
#> ^~~~~~~~~~~

length(BOD):1
#> [1] 2 1

rev(seq_along(BOD))
#> [1] 2 1

^{Created on 2022-09-20 with reprex v2.0.2}

We have a TODO to track this:

lintr/R/seq_linter.R

Lines 67 to 68 in a8d5d1f

	# TODO: better message customization. For example, length(x):1
	# would get rev(seq_along(x)) as the preferred replacement.

[MichaelChirico]

Also for viz: trying to get r-core to give me more confidence in recommending a replacement in the rev() case:

https://bugs.r-project.org/show_bug.cgi?id=18406